log_start_stop() is responsible for logging markers at the beginning and end of a decorated log. They should not be applied if pgit.sh is run with --porcelain. In fact, they are, and vice versa. Fix that.
Set -o pipefail at the start of the script. This makes pgit.sh commit work. Before it didn't, because run_git() doesn't return a proper return value when it's used in a pipe with a cosmetic sed afterwards.
Rename git-show-pushable-master-branches to git-show-pushable, because, the target's recipe doesn't show pushable master branches, but all commits in the working directories current branch not present in origin/master.
Don't use persistent SSH-connections any more for "pkg-" targets,
because, sadly, this hangs after uploading a package.
This commit puts some safeguards against hanging SSH into place, namely setting default SSH timeouts down, SSH keepalive, setting SSH BatchMode to yes.
Use DEP_PROJECTS instead of BUILD_PROJECTS everywhere. DEP_PROJECTS considers more projects, and for all targets it is relevant to, it's desirable to have them operate on the maximum blast radius.
make git-show-pushable-master-branches misses projects which should better be pulled in as dependencies. It searches $(BUILLD_PROJECTS) for repositories with pushable commits, that variable leaves some out, and DEP_PROJECTS has them. It make use of the renovated pkg-requires command.
This move should be extended to other uses of BUILD_PROJECTS as well, after giving it some test runs. And the legacy and redundant command prereq should be removed.
print() should be used to output information requested by a certain command, but not for logging the process to achieve it. log() should be used for the latter. The current code has the distinction not down clearly, fix that.
By default, argcomplete uses argcomplete.FilesCompleter as default for every argument. This mixes accessible files into the list of possible completions. For most of jw-pkg's commands, that's unwanted, so turn it off by defining a NoopCompleter class which does nothing, and by set every arguments's default completer to a NoopCompleter instance. If desired, completing files can be restored for an argument by
Every module derived from lib.Cmd implements its own parser.add_argument() logic. As a consequence, all Cmd derived modules need to be loaded to have the full argument tree available. This is avoided during normal program startup because it takes some time. It's not necessary during normal program execution, nor for showing help messages. It is, however, needed for argcomplete to do its thing, so fully parse the command line if the program runs in argcomplete mode, as determined by checking if _ARGCOMPLETE is present in the environment.
lib.ExecContext.log_delim() logs a header not designed for enclosing command output, and, hence, no footer should be output. This commit suppresses it.
print() should be used to output information requested by a certain command, but not for logging the process to achieve it. log() should be used for the latter. The current code has the distinction not down clearly, fix that.
Log to stderr and add some ASCII-art around the output. Also, add a --porcelain option to allow more stable output parsing. Subsequently, use that option in make targets parsing the output, notably make diff and make git-show-xxx.
Request a remote PTY from AsyncSSH, and wire the local terminal's stdin up with it if interactive == True. This gives a real interactive session if local stdin belongs to a terminal. Also, thanks to AsyncSSH understanding that, forward terminal size changes to the remote end.
Add a SSHClient implementation using AsyncSSH. This is the first and currently only class derived from SSHClient which implements SSHClient.Cap.LogOutput, designed to consume and log command output as it streams in. It felt like the lower hanging fruit not to do that with Paramiko: Paramiko doesn't provide a native async API, so it would need to spawn additional worker threads. I think.
Add an optional caps ("capabilities") argument to the constructor of SSHClient. It is meant to be used by derived classes in order to declare that they don't want the base class to handle a default behaviour for a certain capability, but that they want to implement it themselves instead.
Also, give the _run_ssh() callbacks the necessary info as parameters, so that the derived classes have the means to do so.
Python's platform.system() outputs 'Linux', and to use it is tempting. Sadly, that's wrong, because it reflects the host's idea of the target system, not the execution context's, so replace it with straight 'linux' if the distro is known, or, failing that, the output of uname -s.
Even with --interactive=[true|auto], there's no point in trying to read /etc/os-release interactively, so don't do that. Most notably, this commit keeps the property method from spilling /etc/os-release's content over the terminal.
This commit introduces two new types, Input and InputMode. They replace the more error-prone special strings cmd_input could be used with. InputMode is an Enum, and Input can be either IntputMode, a string or None.
Whether or not the CallContext.interactive property should be True or False, and hence, a call should be processed interactively, depends on multiple factors, constituting matrix of options with multiple preferences.
--interactive is the application default and can be true, false,
or auto
- A call can be explicitly invoked as interactive, non-interactive
or auto via the cmd_input parameter to ExecContext.run()
This commit adds more "mode:" options to make the latter more explicit. It takes preference over the global --interactive parameter: Global --interactive is only given a chance to decide if cmd_input is None (default) or mode:opt-interactive.
This commit also fixes a bug: --interactive is ignored because the interactive argument passed to ExecContext's constructor is ignored later on in calls to the wrapped _run() and _sudo() methods.
Since commit 02697af5, ExecContext.run() returns bytes for stdout and stderr and fixes that in calling code. The thing it did not fix was the code calling run_cmd(), which also made return bytes. This commit catches up on that.
